Value approximation of fuzzy systems variables (Q2277217)

The concept of proximity of fuzzy sets and its measure is introduced. Two fuzzy sets A and \(A':\{x_ 1,x_ 2,...,x_ n\}\to [0,1]\) are said to be approximately equal, \(A\approx A'\), if there exists a small and nonnegative number \(\epsilon\) such that \(\| A-A'\| =\max_{i=1,2,...,n}| A(x_ i)-A'(x_ i)| <\epsilon.\) The number \(\epsilon\) is treated as a proximity index. The basic theorem of the paper shows that the proximity of fuzzy sets under modus ponens realized with the aid of the max-min composition (\(\circ)\) is satisfied, namely \(A\approx A'\) implies that \(A\circ R\approx A'\circ R\). The proximity is not preserved under modus tollens (completed with the use of Godelian implication) applied to determine the upper bound of solutions of the inverse problem in the max-min composition.